porespy
porespy copied to clipboard
Improve ImageJ submodule to run in headless mode
See here for more details.
We've also forced unit tests to only run for Windows. This should also be removed once this issue is dealt with.
For future reference: https://javadoc.scijava.org/ImageJ/net/imagej/ops/image/watershed/Watershed.html
Given that the imagej functions offer only minimal or zero performance improvements, we should probably not invest much more time into this module, and probably deprecate it in v3 down the road.
It seems that the issue has been fixed upstream. Also, it seems that the speedup is real after all, and it's almost an order of magnitude (at least for some filters). So, I'd say let's reopen this issue.
I don't think this is something porespy needs to manage for people. If they want to use imagej funcs from within python, they can figure it out.